Objective: Through the analysis and application of building heating and thermal energy management system, this paper proposes a new thermal energy control strategy to improve the automation level of building heating optimization. Method: This study analyzes the principle of indoor heat balance in buildings. Aiming at the different heating needs of different buildings, a new control strategy is proposed by combining neural network models and fuzzy control theory. Finally, this strategy is applied to the actual building heating, and the practical application value of the strategy proposed by this study is verified through experiments. Result: In the heating stage, after applying the control strategy, the maximum relative error of the temperature is 0.047, and the average error is 0.013. In the antifreeze stage, the maximum error is 0.143 and the average error is 0.09. After the implementation of the control strategy, the temperature fluctuations in the room change little and remain almost between 19 °C and 21 °C. Buildings consume less heat with the highest energy saving rate of 14.37% and the average energy saving rate of 9.23%. Conclusion: The control strategy proposed in this study can adjust the indoor temperature according to the actual situation and achieve the purpose of reasonable heat use. Moreover, it has certain energy-saving effects and can be applied to building heating.